Method and device for segmenting 3D CAD model based on transfinite learning machine

Abstract

The invention discloses a three-dimensional CAD model segmentation method and device based on a transfinite learning machine. The method comprises: computing a feature description operator corresponding to each surface of the three-dimensional CAD model; Training and testing the transfinite learning machine based on the feature descriptor of all surfaces; classifying and labeling every surface of3D CAD model by transfinite learning machine; Based on the classification results, the attribute adjacency markup graph of 3D CAD model being constructed; performing Segmentation of attribute adjacency markup graph; Taking the maximum cohesion of attribute adjacency markup graph segmentation as the objective function, the segmented attribute adjacency markup graph being merged and optimized to obtain multiple local regions. The invention classifies the plane, concave and convex surfaces of a three-dimensional CAD model by a transfinite learning machine, expresses the three-dimensional CAD model by an attribute adjacency mark diagram, and then divides and optimizes the three-dimensional CAD model according to the attribute adjacency mark diagram corresponding to the three-dimensional CAD model.

Description

technical field [0001] The invention relates to the field of three-dimensional CAD model segmentation, in particular to a three-dimensional CAD model segmentation method and device based on an extreme learning machine. Background technique [0002] Segmentation and labeling of CAD models is the basic subject of 3D model feature understanding. Therefore, CAD model segmentation is the key work before the subsequent processing of 3D model features. The traditional surface segmentation method is mainly for mesh models, and the expression of 3D CAD models usually adopts B-rep boundary representation. How to divide the CAD model into local regions with certain engineering significance is an urgent problem to be solved. [0003] Conventional CAD model segmentation is done manually, which has low efficiency and poor precision. With the fast-paced development of the field of machine learning, some new classification methods based on machine learning have been proposed. At present,...

AI Technical Summary

Problems solved by technology

In order to speed up the calculation of clustering, such methods usually first perform over-segmentation on the 3D model, and then perform feature extraction and further clustering, which can greatly improve the speed of the method, but the final segmentation results are heavily dependent on over-segmentation Effect

[0004] To sum up, in the prior art, there is still no effective solution to the problems of long training time, low generalization accuracy and easy to fall into local minimum

Images